Abstract
We present a method for jointly predicting a depth map and intrinsic images from single-image input. The two tasks are formulated in a synergistic manner through a joint conditional random field (CRF) that is solved using a novel convolutional neural network (CNN) architecture, called the joint convolutional neural field (JCNF) model. Tailored to our joint estimation problem, JCNF differs from previous CNNs in its sharing of convolutional activations and layers between networks for each task, its inference in the gradient domain where there exists greater correlation between depth and intrinsic images, and the incorporation of a gradient scale network that learns the confidence of estimated gradients in order to effectively balance them in the solution. This approach is shown to surpass state-of-the-art methods both on single-image depth estimation and on intrinsic image decomposition.
| Original language | English |
|---|---|
| Title of host publication | Computer Vision - 14th European Conference, ECCV 2016, Proceedings |
| Editors | Bastian Leibe, Jiri Matas, Max Welling, Nicu Sebe |
| Publisher | Springer Verlag |
| Pages | 143-159 |
| Number of pages | 17 |
| ISBN (Print) | 9783319464831 |
| DOIs | |
| Publication status | Published - 2016 Jan 1 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 9912 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (Electronic) | 1611-3349 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Computer Science(all)
Cite this
}
Unified depth prediction and intrinsic image decomposition from a single image via joint convolutional neural fields. / Kim, Seungryong; Park, Kihong; Sohn, Kwanghoon; Lin, Stephen.
Computer Vision - 14th European Conference, ECCV 2016, Proceedings. ed. / Bastian Leibe; Jiri Matas; Max Welling; Nicu Sebe. Springer Verlag, 2016. p. 143-159 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9912 LNCS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Unified depth prediction and intrinsic image decomposition from a single image via joint convolutional neural fields
AU - Kim, Seungryong
AU - Park, Kihong
AU - Sohn, Kwanghoon
AU - Lin, Stephen
PY - 2016/1/1
Y1 - 2016/1/1
N2 - We present a method for jointly predicting a depth map and intrinsic images from single-image input. The two tasks are formulated in a synergistic manner through a joint conditional random field (CRF) that is solved using a novel convolutional neural network (CNN) architecture, called the joint convolutional neural field (JCNF) model. Tailored to our joint estimation problem, JCNF differs from previous CNNs in its sharing of convolutional activations and layers between networks for each task, its inference in the gradient domain where there exists greater correlation between depth and intrinsic images, and the incorporation of a gradient scale network that learns the confidence of estimated gradients in order to effectively balance them in the solution. This approach is shown to surpass state-of-the-art methods both on single-image depth estimation and on intrinsic image decomposition.
AB - We present a method for jointly predicting a depth map and intrinsic images from single-image input. The two tasks are formulated in a synergistic manner through a joint conditional random field (CRF) that is solved using a novel convolutional neural network (CNN) architecture, called the joint convolutional neural field (JCNF) model. Tailored to our joint estimation problem, JCNF differs from previous CNNs in its sharing of convolutional activations and layers between networks for each task, its inference in the gradient domain where there exists greater correlation between depth and intrinsic images, and the incorporation of a gradient scale network that learns the confidence of estimated gradients in order to effectively balance them in the solution. This approach is shown to surpass state-of-the-art methods both on single-image depth estimation and on intrinsic image decomposition.
UR - http://www.scopus.com/inward/record.url?scp=84990060400&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84990060400&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-46484-8_9
DO - 10.1007/978-3-319-46484-8_9
M3 - Conference contribution
SN - 9783319464831
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 143
EP - 159
BT - Computer Vision - 14th European Conference, ECCV 2016, Proceedings
A2 - Leibe, Bastian
A2 - Matas, Jiri
A2 - Welling, Max
A2 - Sebe, Nicu
PB - Springer Verlag
ER -